计算 Pandas 数据框中行之间的差异

Question

I have a data frame with data about affiliates, offers, clicks, etc. I want to calculate the difference in conversions( column name « Appr ») between yesterday and today for each offer + affiliate.

D;H;AfID;Affil_name;M;OfID;Offer_name;Clicks;Revenue;Earnings;Appr;Decl;CR;Tr-back
28;11;10;elephant;Ella;1132;App_Aweepstakes;2100;0;0;100;0;1:10;1
28;11;1828;a.kalen;Ella;2675;Cash App_Sweepstakes/CPA_US;3;0;0;200;0;1:50;0
29;11;1828;a.kalen;Ella;2675;Cash App_Sweepstakes/CPA_US;11;0;0;350;0;1:50;0

To do this, I use groupby and diff ():

final_df[´DifAppr’] = final_df.groupby(['H', 'AfID', 'Affil_name', 'M', 'OfID','Offer_name'])[´Appr’].diff().fillna(0)

But if there is no data in the dataframe for the previous day for this offer + affiliate, then this line is ignored and not calculated:

D;H;AfID;Affil_name;M;OfID;Offer_name;Clicks;Revenue;Earnings;Appr;Decl;CR;Tr-back, DiffAppr
29;11;1828;a.kalen;Ella;2675;Cash App_Sweepstakes/CPA_US;11;0;0;350;0;1:50;0;150

I want this line to remain the same in this case. That is, the conversions for the previous day would have been 0, and for the difference in conversions between today and yesterday, the data for today is displayed. That is, for yesterday the affiliate had 0 conversion and this line is not in the dataframe, therefore, today there were 68 conversions. In this case, the "DiffAppr" column should be 68.

Answer 1

I haven't tried it if there are more than two groups in the same group, but if the number of groupings is 1, I don't do anything; if there are more than two, I do a diff() .

final_df['DifAppr'] = (final_df.groupby(['H', 'AfID', 'Affil_name', 'M', 'OfID','Offer_name'])['Appr']
                        .apply(lambda x: x.diff() if len(x) >= 2 else x)).fillna(0)
final_df

D   H   AfID    Affil_name  M   OfID    Offer_name  Clicks  Revenue Earnings    Appr    Decl    CR  Tr-back DifAppr
0   28  11  10  elephant    Ella    1132    App_Aweepstakes 2100    0   0   100 0   1:10    1   100.0
1   28  11  1828    a.kalen Ella    2675    Cash App_Sweepstakes/CPA_US 3   0   0   200 0   1:50    0   0.0
2   29  11  1828    a.kalen Ella    2675    Cash App_Sweepstakes/CPA_US 11  0   0   350 0   1:50    0   150.0